重构主控编排与运行时预热链路，统一工作区提示词/专家调度协议并补齐 wiki 记忆注入与写回闭环。

同时收敛启动与运维脚本默认行为（含 wiki worker）、更新 Admin 可观测性与相关测试，降低首轮时延并提高运行稳定性。

Made-with: Cursor

openclaw/extensions/comfy/image-generation-provider.ts

@@ -0,0 +1,79 @@
+import type {
+  GeneratedImageAsset,
+  ImageGenerationProvider,
+} from "openclaw/plugin-sdk/image-generation";
+import {
+  DEFAULT_COMFY_MODEL,
+  _setComfyFetchGuardForTesting,
+  isComfyCapabilityConfigured,
+  runComfyWorkflow,
+} from "./workflow-runtime.js";
+
+export { _setComfyFetchGuardForTesting };
+
+export function buildComfyImageGenerationProvider(): ImageGenerationProvider {
+  return {
+    id: "comfy",
+    label: "ComfyUI",
+    defaultModel: DEFAULT_COMFY_MODEL,
+    models: [DEFAULT_COMFY_MODEL],
+    isConfigured: ({ cfg, agentDir }) =>
+      isComfyCapabilityConfigured({
+        cfg,
+        agentDir,
+        capability: "image",
+      }),
+    capabilities: {
+      generate: {
+        maxCount: 1,
+        supportsSize: false,
+        supportsAspectRatio: false,
+        supportsResolution: false,
+      },
+      edit: {
+        enabled: true,
+        maxCount: 1,
+        maxInputImages: 1,
+        supportsSize: false,
+        supportsAspectRatio: false,
+        supportsResolution: false,
+      },
+    },
+    async generateImage(req) {
+      if ((req.inputImages?.length ?? 0) > 1) {
+        throw new Error("Comfy image generation currently supports at most one reference image");
+      }
+
+      const result = await runComfyWorkflow({
+        cfg: req.cfg,
+        agentDir: req.agentDir,
+        authStore: req.authStore,
+        prompt: req.prompt,
+        model: req.model,
+        timeoutMs: req.timeoutMs,
+        capability: "image",
+        outputKinds: ["images"],
+        inputImage: req.inputImages?.[0],
+      });
+
+      const images: GeneratedImageAsset[] = result.assets.map((asset) => ({
+        buffer: asset.buffer,
+        mimeType: asset.mimeType,
+        fileName: asset.fileName,
+        metadata: {
+          nodeId: asset.nodeId,
+          promptId: result.promptId,
+        },
+      }));
+
+      return {
+        images,
+        model: result.model,
+        metadata: {
+          promptId: result.promptId,
+          outputNodeIds: result.outputNodeIds,
+        },
+      };
+    },
+  };
+}